Copyright #ERS Journals Ltd 2002

Eur Respir J 2002; 19: 653–657 European Respiratory Journal

DOI: 10.1183/09031936.02.00263102 ISSN 0903-1936
Printed in UK – all rights reserved

Determinants of Fi,O2 with oxygen supplementation during

noninvasive two-level positive pressure ventilation

F. Thys*, G. Liistro#, O. Dozin*, E. Marion*, D.O. Rodenstein#

Determinants of Fi,O2 with oxygen supplementation during noninvasive two-level *Emergency Dept and #Pneumology
positive pressure ventilation. F. Thys, G. Liistro, O. Dozin, E. Marion, D.O. Dept, Cliniques Universitaires Saint-
Rodenstein. #ERS Journals Ltd 2002. Luc, Université Catholique de Louvain,
ABSTRACT: To maintain arterial oxygen saturation (Sa,O2) above 90% in patients Brussels, Belgium.
with acute respiratory failure, oxygen (O2) is often added to the circuit of two-level Correspondence: F. Thys
noninvasive positive pressure ventilation (NPPV). However, the final inspiratory Service des Urgences
oxygen fraction (Fi,O2) is not known. Cliniques Universitaires Saint-Luc
To clarify this issue, the effect of different inspiratory positive airway pressures Université Catholique de Louvain
(IPAP) of the oxygen tubing connection site and the flow rate of O2, on Fi,O2 was Avenue Hippocrate 10
assessed. The effects of the tidal volume (VT) and the respiratory rate on the Fi,O2 were B-1200 Bruxelles
then clarified in a model study. Belgium
The Fi,O2 varied depending on the point where O2 was added to the circuit. When all Fax: 32 27641620
E-mail: Thys@rean.ucl.ac.be
other variables were constant, the connection closest to the exhaust port (ventilator
side) gave the highest Fi,O2. Increases in IPAP led to decreases in Fi,O2. Finally, Fi,O2 Keywords: Inspiratory oxygen fraction
increased with O2 flow, although it was difficult to obtain an Fi,O2 w0.30 unless very noninvasive positive pressure
high O2 flows were used. Paradoxically, NPPV with low IPAP values and without O2 ventilation
supplementation led to a Fi,O2 v0.21 at the circuit-patient interface. VT and respiratory oxygen
rate did not appear to influence Fi,O2.
To conclude, when using noninvasive positive pressure ventilation with two-level respira- Received: July 24 2001
tors, oxygen should be added close to the exhaust port (ventilator side) of the circuit. If ins- Accepted after revision September 10
piratory airway pressure levels arew12 cmH2O, oxygen flows should be at least 4 L?min-1. 2001
Eur Respir J 2002; 19: 653–657.

Currently, two-level noninvasive positive pressure
ventilation (NPPV) is used in the treatment of pati-
ents with acute respiratory failure in intensive care
units [1–7], general pulmonary wards [8, 9] and
emergency depts [10–12]. In these settings, supple-
mental oxygen (O2) is often added to the circuit of
the ventilators to maintain an adequate arterial O2
saturation (Sa,O2). The inspired oxygen fraction (Fi,O2)
is generally unknown, and could be influenced by
a number of factors such as the inspiratory posi-
tive airway pressure (IPAP), the expiratory positive
airway pressure (EPAP), the O2 flow rate and the site
where O2 is added to the circuit etc.
At the present time, there is no published data on
the best way to add O2 to the circuit of a two-level
NPPV (i.e. what level of flow is required, where should
the connection be made). To clarify this matter, a
two-part study was conducted. First, in a clinical
setting, the effect of different IPAP values of the
O2 tubing connection site and the flow rate of O2, on
the Fi,O2 was investigated. Second, an experimental
study to clarify the effect of the tidal volume (VT) and
the respiratory rate on the Fi,O2 was conducted.
Materials and methods
Clinical study
Three normal volunteers, aged 21, 24 and 27 yrs,
in good health and having never smoked were
investigated. Two-level NPPV was initiated with a
barometric ventilator (Bilevel positive airway pres-
sure device (BiPAP1 S/T-D30; Respironics Inc.,
Murrysville, PA, USA), through a face mask (Bird
Corporation, CA, USA), with the subject in a
semirecumbent position. The tubing connecting the
ventilator to the mask had a volume of 565 mL and a
length of 191 cm. The volume between the mask and
the exhaust port as depicted in figure 1 was 19.4 mL.
Initially, the EPAP was set at 2 cmH2O (the minimal
pressure level of the machine) and the IPAP was
set at 2 cmH2O. The IPAP was then increased to 8,
12, 16 and 20 cmH2O. The machine was used in
the assist-control mode with a backup frequency of
12 breaths?min-1 and a back-up inspiratory/expiratory
time ratio of 50%. The device was used with a whisper
swivel-type of exhaust port. Different levels of O2
flow at several different locations in the patient circuit
were added. The O2 was added at the exit of the
BiPAP unit (proximal injection), before the exhaust
port (middle injection) and finally at the patient
connection immediately before the mask (distal injec-
tion) as shown in figure 1. For each level of pressure
and connection point, 0, 2, 4, 6, 8, 10, 12, 14
and 16 L?min-1 of O2 were added. The additional O2
came from a high-pressure source governed by a
flow-meter assembly (Thorpe-tube flow meter) and the
connection was made with a T-connector.
For each value of IPAP, O2 flow rate and
654 F. THYS ET AL.
a)
Expiratory valve Ventilator
A transducer (range ¡5 cmH2O) (Validyne Engineering
Mask D C B Corporation, Northridge, CA, USA) was used for
O2
b)
Expiratory valve Ventilator
A
Mask D C B added before the exhaust port (middle injection in
O2
c)
Expiratory valve Ventilator
A monitor. The BiPAP System was cycled at each test
Mask D C B The final value was the result of three successive
O2
Fig. 1. – Schematic representation of the three types of circuit. a)
proximal injection; b) middle injection; c) distal injection. O2:
oxygen. A: location A, at the exit of the Bilevel positive airway
pressure device; B: B, before the first connector; C: C, between the
two T-connectors; D: D, at the terminal part of the circuit before
the mask.
connection point, the FI,O2 was measured with an O2
monitor (Oxygen monitor 5120; Ohmeda, Madison,
WI, USA). Calibration was carried out according
to the manufacturers9 recommended procedure. The
response time of the O2 monitor was measured at
several different conditions of IPAP and respiratory
rate. The time taken for a 90% change was 43 s
and the time taken for a two-third change was 21 s.
All measurements were performed at a steady state.
The Fi,O2 was measured at 4 points along the circuit
using a second T-connector for each circuit, as
shown in figure 1: location A: at the exit of the
BiPAP unit; location B: before the first T-connector
(exhaust port or O2 addition); location C: between
the two T-connectors; location D: at the terminal part
of the circuit just before the mask. The BiPAP system
was cycled at each test setting until the reading
on the O2 analyser stabilized. The final value was the
result of three successive measures.
Model study
The effects of VT and the respiratory rate were
investigated with a test lung. The test lung was a
bicompartmental model of balloons in a Plexiglas box.
Between the lung model and the ventilator, a low and
a high linear resistance were added to modify the
VT. A number 3 Fleisch pneumotachograph (Fleisch,
Lausanne, Switzerland) with a Validyne pressure
airflow measurement and integration, yielding VT.
Fi,O2, airflow and VT, with the two resistances and
two different rates of ventilation were recorded. The
barometric ventilator used in the clinical study was
also used in this set of experiments.
Initially, the EPAP was set at 2 cmH2O and the
IPAP was set at 2 cmH2O. The IPAP was then
increased to 8, 12, 16 and 20 cmH2O. The barometric
ventilator was used in the controlled mode with two
different frequencies, 14 and 22 breaths?min-1. The
inspiratory/expiratory ratio was 50%. The O2 was
fig. 1). For each level of pressure, 0, 2, 4, 6, 8, 10, 12,
14 and 16 L?min-1 of O2 were added. The additional
O2 came from a high-pressure source governed by a
flow-meter assembly and the connection was made
with a T-connector. For each value IPAP, O2 flow rate
and value of resistance, the Fi,O2 was measured just
before the connection to the lung model with an O2
setting until the reading on the O2 analyser stabilized.
measures.
Statistical analysis
In the clinical part of this study, the relationships
between Fi,O2 and each of the following determinants:
O2 connection point, IPAP, location of measurement
and O2 flow were tested. A standard linear regression
analysis was used. The different relationships obtained
were compared using covariance analysis, in order to
test the difference between their slopes and intercepts.
The same linear regression and covariance analysis
were applied to the variables recorded during the
experimental part of the investigation.
Results
Clinical study
Influence of the oxygen connection point. Figure 2
shows the Fi,O2 values for an IPAP of 8 cmH2O
and an EPAP of 2 cmH2O, according to the three
connection points (proximal, middle, distal injection).
It was shown that, the Fi,O2 changed according to
which connection point was utilized (pv0.05). For
a given O2 flow, the Fi,O2 was greatest when the O2
was connected just before the exhaust port. This
applied to all levels of IPAP and O2 flow and to all
measurement locations.
Influence of the inspiratory positive airway pressure.
Figure 3 shows the Fi,O2 values measured in loca-
tion C with the O2 connected at the middle injection
point, according to each level of IPAP. In this case,
 Fi,O2 DURING NPPV 655

55 50
50
45
45
40
40

Fi,O2 %
Fi,O2 %

35 35
30 30
25
25
20
15 20
1 2 3 4 5 6 7 8 9 4 6 8 10 12 14
Oxygen flow L·min-1 Oxygen L·min-1
Fig. 2. – Trends for the inspiratory oxygen fraction (Fi,O2) values Fig. 4. – Inspiratory oxygen fraction (Fi,O2) values for an inspira-
for an inspiratory positive airway pressure of 8 cmH2O at location tory positive airway pressure (IPAP) of 16 cmH2O with the
C. %: proximal injection; &: middle injection; +: distal injection. oxygen connection at the middle injection for each four measure-
ment locations. %: A; &: B; +: C; h: D. Table 1: Influence of
IPAP on the FI,O2 value in the location C.

55
(pv0.01). This was true for all IPAP, O2 flows and
50 connection points.
45
40 Influence of oxygen flow. When measured at loca-
Fi,O2 %

tion D, with no added O2 and an IPAP v16 cmH2O,

35 the Fi,O2 decreased along the circuit and fell below
30 21%. With an O2 flow of 2 L?min-1, the Fi,O2 decreased
with increasing IPAP levels with measurements
25 performed at location D, but not at the other loca-
20 tions. This applied to all three connection points.
For high IPAP levels (16 and 20 cmH2O), it was
15 difficult to obtain an Fi,O2 w0.35 unless very high O2
0 2 4 6 8 10 12 14 16 flows were used (table 1).
Oxygen L·min-1
Fig. 3. – Inspiratory oxygen fraction (Fi,O2) values at location C
with the oxygen connection at the middle injection for each level Model study
of inspiratory positive airway pressure (IPAP). &: IPAP 8; ':
IPAP 12; h: IPAP 16; #: IPAP 20. For each level of IPAP, VT was lower with a higher
resistance (tables 2 and 3). For a given IPAP and
when levels of O2 flow were low, the value of Fi,O2 resistance, VT decreased with an increase in frequency.
increased as IPAP increased from 8 to 12 cmH2O Under both conditions, it was observed that there was
(pv0.05), but there was a decrease in Fi,O2 when no effect of VT or respiratory frequency on the Fi,O2
pressure was increased from 12 to 16 and 20 cmH2O
(pv0.05). At higher levels of O2 flow, Fi,O2 decreased Table 1. – Influence of inspiratory positive airway pressure
when the IPAP was increased from 8 to 12, 16 (IPAP) on the inspiratory oxygen fraction value in the
and 20 cmH2O. There was no significant difference location C: middle injection
between IPAP of 16 and 20 cmH2O. Therefore in this
circuit, the highest Fi,O2 was obtained with an IPAP of O2 flow L?min-1 IPAP
8 cmH2O. These results did not apply to all circuits,
connection points or O2 flows. For instance, at the 2 8 12 16 20
proximal injection point, when measured at location
C, the Fi,O2 was significantly lower for an IPAP of 0 19 20 21 21 21
20 cmH2O but there was no difference between the 2 26 24 24 24 22
other pressure levels. 4 31 28 29 26 23
6 38 32 33 30 26
8 45 36 36 32 30
Influence of the location of measurement. Figure 4 10 53 40 38 35 35
shows the Fi,O2 values for an IPAP of 16 cmH2O 12 63 44 41 38 37
14 65 47 44 42 39
measured in the four locations, with O2 connected at 16 67 52 48 43 45
the middle injection point. The Fi,O2 increased as the
measurement point was moved closer to the patient Data are presented as %. O2: oxygen.
656 F. THYS ET AL.

Table 2. – Results from the model study at low resistance for the next inspiration. If it was supplied closer to
with 14 breaths?min-1 the mask, the O2 delivered during expiration might
O2 flow L?min-1 Fi,O2 be exhaled through the whisper valve and lost to the
patient.
IPAP 8 IPAP 12 IPAP 16 IPAP 20 The level of IPAP also had an influence on the
Sa,O2. The highest Fi,O2 values into the mask were
0 20 21 21 21
obtained at IPAPs between 8–16 cmH2O. Further-
2 23 26 24 25 more, a drop in the Fi,O2 value was observed in the
4 29 30 30 30 distal part of the circuit with IPAP pressures of
6 37 38 37 35 v8 cmH2O. An Fi,O2 v21% without additional O2
8 46 49 45 42 and IPAP values v16 cmH2O appeared to be indirect
10 67 65 56 48 signs of rebreathing and dilution. Rebreathing pheno-
Tidal volume mL 414 617 797 955 mena have been reported previously with IPAP
Fi,O2: inspiratory oxygen fraction; IPAP: inspiratory positive
v8 cmH2O [13, 14]. At low levels of pressure
airway pressure; O2: oxygen. (v8 cmH2O) and without supplemental O2 the
patients may be submitted to a hypoxic gas mixture.
It could be argued, that the long response time of the
Table 3. – Results from the model study at high resistance O2 monitor produces recordings resulting from the
with 14 breaths?min-1 mixing of inspiratory and expiratory gases. Thus,
the Fi,O2 would be artefactually lowered by the
O2 flow L?min-1 Fi,O2
fractional expired O2. However, this study also reports
IPAP 8 IPAP 12 IPAP 16 IPAP 20 results from the O2 monitor when located at posi-
tion B, where the influence of expired gas would be
minimal. Once O2 was added into the circuit, the Fi,O2
0 20 21 21 21
2 23 25 26 26
decreased with increasing IPAP. This seems logical,
4 28 31 33 32 as higher pressures lead to higher flows of air for a
6 36 38 40 39 fixed flow of O2, probably an unfavourable situation
8 45 46 46 43 for this mixing.
10 60 56 56 51 It is reassuring that in this experimental model,
Tidal volume mL 346 526 707 857 respiratory rate and VT do not affect the Fi,O2.
Although for completion of data collection, the Fi,O2
Fi,O2: inspiratory oxygen fraction; IPAP: inspiratory positive
airway pressure; O2: oxygen.
at different locations was measured, it was clear that
only location D was of clinical relevance, as it was
the location best representing the central inspired
values recorded at different O2 flows, irrespective of O2 concentration at the patient-circuit interface.
IPAP level. Measurements performed within the mask (not used
in this study) may give a better idea of the real
inspired Fi,O2 but the precision and accuracy of
Discussion the O2-sensor cell could be unfavourably influenced
within the mask. Results may also be different with
In the clinical setting, O2 is frequently added to the a nasal mask but the direction of the change would
circuit of two-level NPPV, to maintain Sa,O2 above probably be the same.
90%, in the treatment of patients with acute respira- Conventional ventilators provided with conven-
tory failure. O2 is added to the circuit of the ventilator tional expiratory valves and single tubing still allow
at unspecified points and at different flow rates and some mixing of inspired and expired gases between the
the exact concentration of O2 delivered cannot be airway outlet and the expiratory valve location. These
measured. results could therefore be extended to the conven-
In this investigation, the site of O2 injection appears tional ventilator, although the effect would probably
to be an important factor influencing the concentra- be reduced. By contrast, ventilators provided with
tion of O2 that the patient receives. The highest values separate expiratory and inspiratory tube lines should
of Fi,O2 were recorded when the O2 was introduced not have this problem, although this remains to be
to the circuit just before the expiratory port. Connec- tested.
ting O2 closer to the respirator or closer to the patient A caveat concerning the absence of the influence
resulted in reduced values of Fi,O2 for the same O2 of VT on Fi,O2 seems necessary. Indeed the potential
flows and ventilator settings. The mixture of air and role of VT on a model lung, where there was neither
O2 is probably more homogeneous when injected O2 consumption nor carbon dioxide production, was
in the middle than in the proximal or distal loca- assessed. One might suppose that VT could behave
tions. It may be hypothesized that if the O2 flow is in a more disturbing way in a patient, by its influence
added between the mask and the expiratory port, on dead space and, eventually, rebreathing into the
the blending of the expiratory and inspiratory gases distal part of the circuit. Nevertheless, the model
could lower the Fi,O2 on the patient9s side. O2 was data in this study (VT changing two-fold) suggests
delivered continuously and although it was not that, the problem should be slight.
studied, there is a possibility that if O2 was supplied To the best of the authors9 knowledge, this is the
just before the whisper valve it may form a reservoir first study examining the determinants of Fi,O2 during
 Fi,O2 DURING NPPV
NPPV. This is an incomplete study. Indeed measure-
ments using one model of two-level NPPV, only one
connecting tube and a single face mask have been
performed. The subjects studied were normal subjects,
with normal lung mechanics and normal dead spaces.
Moreover, the possible effect of change in EPAP
on Fi,O2 has not been investigated. It is clear that
these results could be different if measurements were
performed with different respirators, different tubing
lengths and volumes, and in patients with different
lung diseases. Nevertheless, the important point still
remains that Fi,O2 depends on several determinants in
addition to O2 flow.
To conclude, when using supplemental oxygen
during noninvasive positive pressure ventilation, the
inspiratory oxygen fraction depended on three major
factors: the point where oxygen is added into the
circuit, the level of inspiratory positive airway pres-
sure, and the oxygen flow rate. The respiratory rate
and the tidal volume did not influence the delivered
inspiratory oxygen fration. For inspiratory positive
airway pressures w12 cmH2O, the inspiratory oxygen
fraction flows should be at least 4 L?min-1. An
inspiratory oxygen fraction o0.5 requires a very
high level of oxygen flow. This was a limited
experimental study, and should be considered as a
guide rather than a complete predictor for different
two-level pressure support ventilators used with
various masks or different levels of expiratory positive
airway pressures.
Acknowledgements. The authors would like
to thank O. Pitance and D. Reychler for their
help in collecting the data and N. Stroobant
for her constant dedication to this study.
References
1. Meyer TJ, Hill NS. Non invasive positive pressure
ventilation to treat respiratory failure. Ann Intern Med
1994; 120: 760–770.
2. Freichels TA. Palliative ventilatory support: Use of
non invasive positive pressure ventilation in terminal
respiratory insufficiency. Am J Crit Care 1994; 3: 6–10.
657
3. Pennock BE, Crawshaw L, Kaplan PD. Non invasive
nasal mask ventilation for acute respiratory failure:
Institution of a new therapeutic technology for routine
use. Chest 1994; 105: 441–444.
4. Wysocki M, Tric L, Wolff MA, Gertner J, Millet H,
Herman B. Non invasive pressure support ventilation
in patients with acute respiratory failure. Chest 1993;
103: 907–913.
5. Foglio C, Vittaca M, Quadri A, Scalvini S, Marangoni
S, Ambrosino N. Acute exacerbation in COLD
patients: Treatment using positive pressure ventilation
by nasal mask. Chest 1992; 101: 1533–1538.
6. Meduri GU. Noninvasive positive pressure ventilation
in patients with acute respiratory failure. Clin Chest
Med 1996; 17: 513–533.
7. Meduri GU, Abou-Shala N, Fox RC, Jones CB,
Leeper KV, Wunderink RG. Noninvasive face mask
mechanical ventilation in patients with acute hyper-
capnic respiratory failure. Chest 1991; 100: 445–454.
8. Barbe F, Togores B, Rubi M, Pons S, Maimo A,
Agusti AG. Noninvasive ventilatory support does not
facilitate recovery from acute respiratory pulmonary
disease. Eur Respir J 1996; 9: 1240–1245.
9. Bott J, Carroll M, Conway J, et al. A randomized-
controlled study of nasal intermittent positive pressure
ventilation in acute exacerbations of chronic obstruc-
tive airways disease. Lancet 1993; 341: 1555–1557.
10. Pollack CV, Torres MT, Alexander L. Feasibility
study of the use of bilevel positive airway pressure
for respiratory support in emergency department. Ann
Emerg Med 1996; 27: 189–192.
11. Celikel T, Sungur M, Ceyhan B, Karakurt S.
Comparison of noninvasive positive pressure ventila-
tion with standard medical therapy in hypercapnic
acute respiratory failure. Chest 1998; 114: 1636–1642.
12. Thys F, Roeseler J, Delaere S, et al. Two-level non
invasive positive pressure ventilation in the initial
treatment of acute respiratory failure in an emergency
department. Eur J Emerg Med 1999; 6: 207–214.
13. Ferguson GT, Gilmartin M. CO2 rebreathing during
BiPAP ventilatory assistance. Am J Respir Crit Care
Med 1995; 151: 1126–1135.
14. Lofaso F, Brochard L, Touchard D, Hang T, Harf A,
Isabay D. Evaluation of carbon dioxyde rebreath-
ing during pressure support ventilation with airway
management system (BiPAP) devices. Chest 1995; 108:
772–78.